Cumulative probabilities of live birth across multiple complete IVF/ICSI cycles: a call for attention

Predicting the number of oocytes retrieved from controlled ovarian hyperstimulation with machine learning

STUDY QUESTION

Can machine learning predict the number of oocytes retrieved from controlled ovarian hyperstimulation (COH)?

SUMMARY ANSWER

Three machine-learning models were successfully trained to predict the number of oocytes retrieved from COH.

WHAT IS KNOWN ALREADY

A number of previous studies have identified and built predictive models on factors that influence the number of oocytes retrieved during COH. Many of these studies are, however, limited in the fact that they only consider a small number of variables in isolation.

STUDY DESIGN, SIZE, DURATION

This study was a retrospective analysis of a dataset of 11,286 cycles performed at a single centre in France between 2009 and 2020 with the aim of building a predictive model for the number of oocytes retrieved from ovarian stimulation. The analysis was carried out by a data analysis team external to the centre using the Substra framework. The Substra framework enabled the data analysis team to send computer code to run securely on the centre's on-premises server. In this way, a high level of data security was achieved as the data analysis team did not have direct access to the data, nor did the data leave the centre at any point during the study.

PARTICIPANTS/MATERIALS, SETTING, METHODS

The Light Gradient Boosting Machine algorithm was used to produce three predictive models: one that directly predicted the number of oocytes retrieved and two that predicted which of a set of bins provided by two clinicians the number of oocytes retrieved fell into. The resulting models were evaluated on a held-out test set and compared to linear and logistic regression baselines. In addition, the models themselves were analysed to identify the parameters that had the biggest impact on their predictions.

MAIN RESULTS AND THE ROLE OF CHANCE

On average, the model that directly predicted the number of oocytes retrieved deviated from the ground truth by 4.21 oocytes. The model that predicted the first clinician's bins deviated by 0.73 bins whereas the model for the second clinician deviated by 0.62 bins. For all models, performance was best within the first and third quartiles of the target variable, with the model underpredicting extreme values of the target variable (no oocytes and large numbers of oocytes retrieved). Nevertheless, the erroneous predictions made for these extreme cases were still within the vicinity of the true value. Overall, all three models agreed on the importance of each feature which was estimated using Shapley Additive Explanation (SHAP) values. The feature with the highest mean absolute SHAP value (and thus the highest importance) was the antral follicle count, followed by basal AMH and FSH. Of the other hormonal features, basal TSH, LH, and testosterone levels were similarly important and baseline LH was the least important. The treatment characteristic with the highest SHAP value was the initial dose of gonadotropins.

LIMITATIONS, REASONS FOR CAUTION

The models produced in this study were trained on a cohort from a single centre. They should thus not be used in clinical practice until trained and evaluated on a larger cohort more representative of the general population.

WIDER IMPLICATIONS OF FINDINGS

These predictive models for the number of oocytes retrieved from COH may be useful in clinical practice, assisting clinicians in optimizing COH protocols for individual patients. Our work also demonstrates the promise of using the Substra framework for allowing external researchers to provide clinically relevant insights on sensitive fertility data in a fully secure, trustworthy manner and opens a number of exciting avenues for accelerating future research.

STUDY FUNDING/COMPETING INTEREST(S)

This study was funded by the French Public Bank of Investment as part of the Healthchain Consortium. T.Fe., C.He., J.C., C.J., C.-A.P., and C.Hi. are employed by Apricity. C.Hi. has received consulting fees and honoraria from Vitrolife, Merck Serono, Ferring, Cooper Surgical, Dibimed, Apricity, and Fairtility and travel support from Fairtility and Vitrolife, participates on an advisory board for Merck Serono, was the founder and organizer of the AI Fertility conference, has stock in Aria Fertility, TMRW, Fairtility, Apricity, and IVF Professionals, and received free equipment from Planar in exchange for first user feedback. C.J. has received a grant from BPI. J.C. has also received a grant from BPI, is a member of the Merck AI advisory board, and is a board member of Labelia Labs. C.He has a contract for medical writing of this manuscript by CHU Nantes and has received travel support from Apricity. A.R. haș received honoraria from Ferring and Organon. T.Fe. has received a grant from BPI.

TRIAL REGISTRATION NUMBER

N/A.

OUP accepted manuscript

BACKGROUND

Oocyte activation deficiency (OAD) is attributed to the majority of cases underlying failure of ICSI cycles, the standard treatment for male factor infertility. Oocyte activation encompasses a series of concerted events, triggered by sperm-specific phospholipase C zeta (PLCζ), which elicits increases in free cytoplasmic calcium (Ca²⁺) in spatially and temporally specific oscillations. Defects in this specific pattern of Ca²⁺ release are directly attributable to most cases of OAD. Ca²⁺ release can be clinically mediated via assisted oocyte activation (AOA), a combination of mechanical, electrical and/or chemical stimuli which artificially promote an increase in the levels of intra-cytoplasmic Ca²⁺. However, concerns regarding safety and efficacy underlie potential risks that must be addressed before such methods can be safely widely used.

OBJECTIVE AND RATIONALE

Recent advances in current AOA techniques warrant a review of the safety and efficacy of these practices, to determine the extent to which AOA may be implemented in the clinic. Importantly, the primary challenges to obtaining data on the safety and efficacy of AOA must be determined. Such questions require urgent attention before widespread clinical utilization of such protocols can be advocated.

SEARCH METHODS

A literature review was performed using databases including PubMed, Web of Science, Medline, etc. using AOA, OAD, calcium ionophores, ICSI, PLCζ, oocyte activation, failed fertilization and fertilization failure as keywords. Relevant articles published until June 2019 were analysed and included in the review, with an emphasis on studies assessing large-scale efficacy and safety.

OUTCOMES

Contradictory studies on the safety and efficacy of AOA do not yet allow for the establishment of AOA as standard practice in the clinic. Heterogeneity in study methodology, inconsistent sample inclusion criteria, non-standardized outcome assessments, restricted sample size and animal model limitations render AOA strictly experimental. The main scientific concern impeding AOA utilization in the clinic is the non-physiological method of Ca²⁺ release mediated by most AOA agents, coupled with a lack of holistic understanding regarding the physiological mechanism(s) underlying Ca²⁺ release at oocyte activation.

LIMITATIONS, REASONS FOR CAUTION

The number of studies with clinical relevance using AOA remains significantly low. A much wider range of studies examining outcomes using multiple AOA agents are required.

WIDER IMPLICATIONS

In addition to addressing the five main challenges of studies assessing AOA safety and efficacy, more standardized, large-scale, multi-centre studies of AOA, as well as long-term follow-up studies of children born from AOA, would provide evidence for establishing AOA as a treatment for infertility. The delivery of an activating agent that can more accurately recapitulate physiological fertilization, such as recombinant PLCζ, is a promising prospect for the future of AOA. Further to PLCζ, many other avenues of physiological oocyte activation also require urgent investigation to assess other potential physiological avenues of AOA.

STUDY FUNDING/COMPETING INTERESTS

D.G. was supported by Stanford University’s Bing Overseas Study Program. J.K. was supported by a Healthcare Research Fellowship Award (HF-14-16) made by Health and Care Research Wales (HCRW), alongside a National Science, Technology, and Innovation plan (NSTIP) project grant (15-MED4186-20) awarded by the King Abdulaziz City for Science and Technology (KACST). The authors have no competing interests to declare.

Predictors of live birth in women with a history of biochemical pregnancies after assisted reproduction treatment

OBJECTIVES

To ascertain whether women with a history of biochemical pregnancies (BPs) in in-vitro fertilization (IVF)/intracytoplasmic sperm injection (ICSI) treatment cycles have decreased chances of live birth (LB); and (2) to build a predictive model for LB in this particular population of infertile women.

METHODS

In order to achieve the first objective, data from 1536 women that had a LB using autologous fresh or frozen embryos, or dropped out of further IVF/ICSI treatments after completing one to three unsuccessful treatment cycles were retrospectively analyzed. A subpopulation of 90 women that experienced one or more BPs in our assisted reproduction unit were selected to build a predictive logistic regression model for LB.

RESULTS

LB percentages significantly decreased from a value of 55.3 % in women with no history of previous BPs to 30.9 % and 11.1 % in women that displayed a history of one or more than one BP, respectively. Three out of 35 selected potential predictors were finally included into the model: "number of the last embryo transfer cycle resulting in a BP", "women's age", and "oligo-, astheno-, and/or teratozoospermia". The value of the c-statistic was 0.819 (asymptotic 95 % CI: 0.724-0.913). The model adequately fitted the data with no significant over or underestimation of predictor effects.

CONCLUSION

(1) A history of BPs is negatively associated with later chance of LB in women undergoing a series of IVF/ICSI treatment cycles; and (2) LB probability of women with a history of BPs can be predicted using a model with excellent discriminatory capacity.

A predictive model for women's assisted fecundity before starting the first IVF/ICSI treatment cycle

PURPOSE

To introduce a prognostic model for women's assisted fecundity before starting the first IVF/ICSI treatment cycle.

METHODS

In contrast to previous predictive models, we analyze two groups of women at the extremes of prognosis. Specifically, 708 infertile women that had either a live birth (LB) event in the first autologous IVF/ICSI cycle ("high-assisted-fecundity women", n = 458) or did not succeed in having a LB event after completing three autologous IVF/ICSI cycles ("low-assisted-fecundity women", n = 250). The initial sample of 708 women was split into two sets in order to develop (n = 531) and internally validate (n = 177) a predictive logistic regression model using a forward-stepwise variable selection.

RESULTS

Seven out of 32 initially selected potential predictors were included into the model: women's age, presence of multiple female infertility factors, number of antral follicles, women's tobacco smoking, occurrence of irregular menstrual cycles, and basal levels of prolactin and LH. The value of the c-statistic was 0.718 (asymptotic 95% CI 0.672-0.763) in the development set and 0.649 (asymptotic 95% CI: 0.560-0.738) in the validation set. The model adequately fitted the data with no significant over or underestimation of predictor effects.

CONCLUSION

Women's assisted fecundity may be predicted using a relatively small number of predictors. This approach may complement the traditional procedure of estimating cumulative and cycle-specific probabilities of LB across multiple complete IVF/ICSI cycles. In addition, it provides an easy-to-apply methodology for fertility clinics to develop and actualize their own predictive models.